Abstract: Common methods of reservoir fracture identification and prediction include conventional well logging and FMI logging technologies, which are mainly applied to carbonate and sandstone reservoirs. However, less studies were carried out, and no effective method for fracture identification and prediction of sandy conglomerate reservoirs with ultra-low permeability. In this study, taking the sandy conglomerate reservoir with ultra-low permeability in Well Hong 153 in the Junggar Basin as an example, a quantitative characterization method has been established, which integrates geological, seismic and well logging techniques. First, the causes, development characteristics and main controlling factors of fractures are analyzed by cores, thin sections and other data to describe fractures at macroscope scale; second, fracture identification and interpretation models are built using core, FMI logging and conventional logging data, which are used for single-well fracture identification and prediction of sandy conglomerate reservoirs with ultra-low permeability; third, after extracting prestack seismic anisotropic attributes, fracture density and orientation are predicted based on seismic, well and sedimentary data; finally fracture identification and prediction results are verified by core, FMI logging, well and production test data. Application of the method to Well Hong 153 shows that the accuracy of fracture identification is 71.3%, and proves that the fracture prediction results from pre-stack seismic data are in good agreement with the actual production conditions.

Key words: ultra-low permeability, sandy conglomerate reservoir, fracture identification, prestack seismic prediction, model method